Researchers at Michigan State University are leveraging DeepMind’s AlphaFold to redesign a photosynthetic enzyme that falters under heat. Associate professor Berkley Walker targets glycerate kinase, a protein that recycles carbon during photosynthesis. Rising global temperatures threaten this molecular choreography, shrinking yields of staple crops worldwide in future.

Because no experimental structure existed for plant glycerate kinase, Walker’s team fed AlphaFold predicted models of both the plant enzyme and a thermophilic algae counterpart into molecular dynamics simulations. The simulations revealed three flexible loops in the plant version that unraveled at elevated temperatures, while the algae loops remained rigid.

Armed with those insights, the lab engineered hybrid enzymes by swapping the unstable plant loops for the sturdier algae segments. One variant retained activity up to 65 °C, a temperature far beyond typical field conditions. Laboratory assays confirmed that the chimeric glycerate kinase preserved carbon‑recycling function where the native protein failed.

This proof‑of‑concept shows how AI‑predicted structures can shortcut years of trial‑and‑error protein engineering, directly feeding into breeding programs that aim for climate‑resilient crops. By stabilizing a single enzyme, researchers open a pathway to sustain photosynthetic efficiency even as heatwaves become more frequent globally today again.